WIP | Tico Me

About

This electron app uses whisper to transcribe audio and uploads the transcript to the Voiceflow Knowledge Base. This has been tested on MacOs and might need some adjustments to run on other platforms.

Config

Rename the config.json.template to config.json and update the values.

If you have access to the KB Table (JSON) BETA, set kbJSON to true otherwise, set the voiceflowProjectId value to your project's ID and set kbJSON to false

Whisper

For this demo, we are using the following project to run whisper as a webservice:

Github repo: https://github.com/ahmetoner/whisper-asr-webservice

Doc: https://ahmetoner.github.io/whisper-asr-webservice

To run the whisper webservice, pull the docker image and run the container.

docker pull onerahmet/openai-whisper-asr-webservice:latest
docker run -d --restart unless-stopped -p 9000:9000 -e ASR_MODEL=base.en
-e ASR_ENGINE=faster_whisper onerahmet/openai-whisper-asr-webservice:latest

Set whisperUrl in the config.json file to the url of the whisper instance.

Setup

npm install

The node-record-lpcm-16 module requires you to install SoX and it must be available in your $PATH.

We've added a post-install script to check for SoX. If SoX is not installed, this script will try to install it for you. If for any reason this script doesn't work, you can try to install SoX manually:

For Mac OS brew install sox

For most linux disto's sudo apt-get install sox libsox-fmt-all

For Windows Working version for Windows is 14.4.1. You can download the binaries or use chocolately to install the package

choco install sox.portable

Run the app

npm start

Build the app (tested on Mac only)

You will have to edit the package.json file build to set the correct values.

npm run build